CREB-Binding Protein Is a Mediator of Neuroblastoma Cell Death Induced by the Histone Deacetylase Inhibitor Trichostatin A1

RESEARCH ARTICLE Neoplasia . Vol. 9, No. 6, June 2007, pp. 495 – 503 495 www.neoplasia.com

CREB-Binding Protein Is a Mediator of Neuroblastoma Cell Death Induced By the Histone Deacetylase Inhibitor Trichostatin A1

Chitra Subramanian*, Jason A. Jarzembowski y,2, Anthony W. Opipari Jr. z, Valerie P. Castle* and Roland P. S. Kwok z,§

Departments of *Pediatrics, yPathology, zObstetrics and Gynecology, and §Biological Chemistry, University of Michigan, Ann Arbor, MI 48109, USA

Abstract The cytotoxic mechanism of the histone deacetylase involved in the regulation of nearly every major cellular pro- inhibitor (HDACI) Trichostatin A (TSA) was explored in a cess. As such, activators and inhibitors of HATs and HDACs neuroblastoma (NB) model. TSA induces cell death in are important to understanding molecular regulation and are neuroblastic-type NB cells by increasing the acetyla- of increasing interest as therapeutic agents. For example, in- tion of Ku70, a Bax-binding protein. Ku70 acetylation hibitors of HDACs have been approved for clinical use as anti- causes Bax release and activation, triggering cell death. tumor agents [2]. Despite evidence supporting their efficacy in This response to TSA depends on the CREB-binding several types of cancer, molecular mechanisms linking HDAC protein (CBP) acetylating Ku70. TSA-induced cell death inhibition to effects on cell survival and growth are not under- response correlates with CBP expression. In stromal- stood. Consequently, the HDAC sensitivity or resistance of type NB cell lines with low levels of CBP and relative tumors cannot be rationally predicted. resistance to TSA, increasing CBP expression disrupts Because histones and transcription factors were the first Bax–Ku70 binding and sensitizes them to TSA. Re- proteins to be identified as HDAC substrates, gene transcription ducing CBP expression in neuroblastic cell types and chromatin remodeling were the initial foci of experiments causes resistance. Cytotoxic response to TSA is Bax- studying the antitumor mechanisms of histone deacetylase in- dependent. Interestingly, depleting NB cells of Ku70 hibitors (HDACIs) [3]. However, we have provided strong evi- also triggers Bax-dependent cell death, suggesting that dence for a non–chromatin-based mechanism, involving Ku70 conditions that leave Bax unbound to Ku70 result in protein, that accounts for antitumor proapoptotic response in cell death. We also show that CBP, Ku70, and Bax are neuroblastoma (NB) tumor cells [4]. In neuroblastic NB cell expressed in human NB tumors and that CBP expres- lines (N-type) analogous to the predominant tumor cell type in sion varies across cell types comprising these tumors, NB tumors, about 50% of Ku70 protein is localized in the cytosol with the highest expression observed in neuroblastic where it is bound to Bax, a multidomain proapoptotic member elements. Together, these results demonstrate that of the Bcl-2 family of proteins [5]. Ku70 was originally charac- CBP, Bax, and Ku70 contribute to a therapeutic re- terized as an autoantigen protein [6,7], which is essential for sponse to TSA against NB and identify the possibility the repair of nonhomologous DNA double-strand breaks as of using these proteins to predict clinical responsive- part of the DNA-binding component of DNA-dependent pro- ness to HDACI treatment. tein kinase [8,9]. Distinct from this biochemical activity, Ku70 Neoplasia (2007) 9, 495–503 regulates tumor cell survival through its association with Bax [10]. The carboxy-terminal region of Ku70, which binds Bax, Keywords: Histone acetyltransferase, histone deacetylase, Ku70, Bax, apoptosis. contains a domain that interacts with HATs (CBP and p300/

Abbreviations: CBP, CREB-binding protein; HAT, histone acetyltransferase; HDACI, histone deacetylase inhibitor; NB, neuroblastoma; TSA, Trichostatin A Introduction Address all correspondence to: Roland Kwok, 6428 Medical Science I, 1150 West Medical Protein function can be regulated by posttranslational modi- Center Drive, Ann Arbor, MI 48109-0617. E-mail: [email protected] 1R.P.S.K. was supported by National Institutes of Health grant DK067102; A.W.O. and V.P.C. fications, including phosphorylation, sumoylation, ubiquiti- were supported by the Ravitz Foundation; and V.P.C. was supported by the Janette Ferrantino nation, and acetylation. The acetylation status of proteins is Hematology Research Fund. This work used the Sequencing Core of the Michigan Diabetes Research and Training Center, which was funded by National Institute of Diabetes and regulated by two groups of enzymes: histone acetyltrans- Digestive and Kidney Diseases grant NIH5P60 DK20572. 2 ferases (HATs), including CREB-binding protein (CBP) and Current address: Department of Pathology, Washington University Medical Center, St. Louis, MO 63110. p300, and histone deacetylases (HDACs). Despite the no- Received 15 March 2007; Revised 24 April 2007; Accepted 27 April 2007. menclature, both HATs and HDACs act on nonhistone pro- Copyright D 2007 Neoplasia Press, Inc. All rights reserved 1522-8002/07/$25.00 teins as well as on histones [1], and protein acetylation is DOI 10.1593/neo.07262 496 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al.

CBP–associated factor) [10]. Specific residues in this domain transfection, SH-EP1 cells were transfected with the vector can be acetylated (lysine-539 and lysine-542), preventing or pcDNA3-CBP-FLAG or pcDNA3-p300-FLAG, with or without disrupting Bax binding. Treating cells with Trichostatin A (TSA; pCMV2B-Ku70 wild-type FLAG or pCMV2B-Ku70 K539/ an inhibitor of class I/II HDACs) or nicotinamide (a class III 542R-FLAG, and a green fluorescence protein (GFP) expres- HDACI) increases the acetylation of these residues [10], dis- sion vector (Amaxa, Gaithersburg, MD) using Nucleofector rupts the Ku70–Bax complex, and sends Bax to the mito- kit V (Amaxa). Briefly, immediately before transfection, 2 chondria to trigger apoptotic cell death. Thus, it appears that 106 cells were trypsinized, washed with phosphate-buffered the balance of acetylation on these key regulatory sites is, at saline, and then resuspended in Nucleofector V solution least partially, determined by HDACs [11]. Moreover, high containing 5 mg of indicated plasmids along with 0.5 mgof levels of Ku70 expression are observed in N-type cell lines. the GFP vector to a final concentration of 3 106 to 4 Although this may be a factor that limits cellular sensitivity to 106 cells/100 ml. Samples were subjected to nucleofection radiation and certain types of chemotherapy, the cellular using the Nucleofector device at settings recommended by sensitivity of N-type cell lines to HDACIs appears instead to the manufacturer for SH-SY5Y cells (Amaxa). After transfec- be directly proportional to Ku70 expression. tion, cells were resuspended in prewarmed MEM and plated. In addition to neuroblastic cells, NB tumors contain a Twenty-four hours after transfection, the efficiency of trans- second tumor-derived or tumor-associated cell population, fection was determined by GFP positivity. stromal-type (S-type) cells. Whereas N-type cell lines express N-myc protein and neuronal markers, represent a Cell Viability Assays larger fraction of total tumor cells in high-risk tumors, and TwoS-type(SH-EP1andSK-N-AS)andtwoN-type are tumorigenic in xenografts, S-type cell lines do not ex- (IMR32 and SH-SY5Y) human NB cell lines were treated with press N-myc, express stromal cell markers, are not tumori- different concentrations of TSA, and viability was determined genic in mice, and are associated with less malignant tumor after 24 and 48 hours by MTT assay, as previously described behavior. We conducted experiments to determine if the [4]. The viabilities of CBP-transfected and CBP knockdown response of S-type NB tumor cell lines to TSA was similar cells were similarly measured. All experiments were per- to that of N-type cell lines. Interestingly, S-type NB cells are formed at least three times, with triplicates in each experi- significantly less sensitive to this HDACI. This lack of re- ment. Data from a representative experiment are shown as sponsiveness is seen both with overall cell survival and at the mean values and standard deviations. mechanistic level of disruption of the Bax–Ku70 complex. Comparing S-type cell lines with N-type cell lines showed that CBP is differentially expressed (higher in N-type than siRNA-Mediated Silencing S-type) so as to potentially account for differential respon- Briefly, 24 hours before transfection, SH-SY5Y cells were 6 siveness. When CBP expression was increased in S-type plated at a density of 2 10 cells per 10-cm plate. On the fol- cell lines, TSA treatment results in cell death mediated by lowing day, cells were transfected with SMART-pool CBP Bax release that depends on acetylation-sensitive sites in siRNA (Dharmacon, Inc., Lafayette, CO) using a Nucleofector Ku70. When CBP expression in N-type cell lines was re- transfection reagent (Amaxa), as per the manufacturer’s pro- duced, cells became relatively resistant to TSA treatment. tocol. As a measure of transfection efficiency, we used siGlo Reducing Bax expression in these cells, like decreasing (Dharmacon). Mock transfection and transfection with scram- CBP, similarly reduced their sensitivity to treatment with this bled siRNA served as controls. siRNA-induced depletion of HDACI. Conversely, reducing Ku70 levels in these cells CBP expression was measured 48 to 72 hours after transfec- caused cell death even without HDACI treatment. This was tion by Western blot analysis using CBP antibodies (06-893; likewise dependent on Bax expression. Together with data Santa Cruz Biotechnology, Inc., Santa Cruz, CA). presented on the expression patterns of CBP, Ku70, and Bax in a collection of human NB tumor specimens, our re- Western Blot Analysis and Coimmunoprecipitation sults showing that these proteins together mediate cell death Whole-cell extracts of NB cell lines SH-SY5Y, IMR32, SH- in response to TSA in NB cells provide evidence that these EP1, SK-N-AS, GOTO, SMS-KCN-69n, LA1-55n LA1-5s, proteins are potential biomarkers for predicting and moni- SH-310, and WSN were separated by sodium dodecyl toring treatment responsiveness to HDACIs in NB tumors. sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), transferred to PDVF membrane, and then immunoblotted for CBP, p300, Bax, or Ku70. Tubulin was used as a loading Materials and Methods control. The following antibodies were used for Western blot analyses: tubulin (T-4026) and FLAG (F-1804) antibodies Cell Culture and Cell Transfection were from Sigma (St. Louis, MO); Ku70 (A-9), CBP (A-22), Human NB cell lines SH-SY5Y,IMR32, SH-EP1, SK-N-AS, and p300 (N-15) antibodies were from Santa Cruz Bio- GOTO, SMS-KCN-69n, LA1-55n LA1-5s, SH-310, and WSN chemicals; and Bax-NT (06-499) antibody was from Upstate were cultured in modified Eagle’s medium (MEM) sup- (Billerica, MA). Western blot analyses were visualized by using plemented with 10% fetal bovine serum, 100 U/ml penicillin, Enhanced Chemiluminescence Plus (Amersham Pharmacia and 100 mg/ml streptomycin. The cells were maintained Biotech, Piscataway, NJ). Coimmunoprecipitation of endoge- at 37jC in a humidified 5% CO2 incubator. For transient nous and FLAG-tagged Ku70 before and after TSA treatment

Neoplasia . Vol. 9, No. 6, 2007 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al. 497 was performed in CHAPS buffer, according to the procedure Results described by Sawada et al. [5]. S-Type NB Cells Are Less Sensitive to TSA and Have Lower NB Tissue Array Levels of HATs Than N-Type Cells Growing in Culture Tissue arrays were constructed using triplicate 1.0-mm To determine the relative responsiveness of N-type and cores taken from 45 paraffin-embedded formalin-fixed S-type NB cell lines to TSA-induced cell death, we treated neuroblastic tumors (30 NBs, 7 ganglioneuroblastomas, 7 two N-type (IMR32 and SH-SY5Y) and two S-type (SH-EP1 ganglioneuromas, and 5 composite/mixed-histology tumors), and SK-N-AS) NB cell lines with TSA for 24 or 48 hours, and 5 normal adrenal glands, and 20 other neural crest–derived then measured cell viability with MTT assay. After 48 hours, and unrelated neoplasms. After constructing and curing 1 mM TSA reduced the viability of N-type cells to less than array cores in a paraffin block, 10-mm sections on poly-L- 10% to 20% (Figure 1A). In contrast, S-type cells were sig- lysine slides were taken for routine hematoxylin–eosin stain- nificantly less sensitive to TSA: at 1 mM, viability remained ing and immunohistochemistry. Staining was performed above 50% (Figure 1B). Consistent with our previous report, according to the manufacturer’s recommended Immuno- TSA disrupts the molecular interaction between Ku70 and histochemistry (IHC) protocols using the same antibodies Bax in N-type SH-SY5Y (Figure 2A, lanes 5 and 6) and described above. A semiquantitative system was used to IMR32 cells (data not shown), which we proposed is due to measure staining intensity, with a score from 0 to 3 assigned increased acetylation on Ku70 lysine residues K539 and to each component (neuroblastic and stromal) of each tumor K542 [4]. In contrast, although SH-EP1 cells have similar core. Mean scores and component ratios (neuroblast stain- levels of Ku70–Bax complex detected in cytosolic extracts ing intensity divided by stromal cell staining intensity) were (Figure 2A, lanes 1–4), TSA does not affect Bax and Ku70 then derived for each tumor from its triplicate cores. Corre- binding in these cells (Figure 2A, lanes 7 and 8). These data lations between immunohistochemical staining and histo- are generally consistent with findings recently reported by logic cell type were performed using bivariate analysis with Furchert et al. [12] comparing growth inhibition in response Pearson coefficients and two-tailed significance tests. to TSA and other HDACIs in a panel of cells, including five

Figure 1. TSA induces cell death in N-type, but not S-type, NB cells. N-type cell lines (IMR32 and SH-SY5Y) (A) and S-type cell lines (SH-EP1 and SK-N-AS) (B) were treated with various concentrations of TSA for 24 or 48 hours, as indicated. Cell viability was determined by MTT assay. Results are expressed as the percentage of viable cells compared with vehicle-treated controls (mean ± SD; n = 3).

Neoplasia . Vol. 9, No. 6, 2007 498 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al.

Figure 2. N-type NB cells have higher CBP and p300 levels, and Bax is released from Ku70 in response to TSA. (A) Both N-type (SH-SY5Y) and S-type (SH-EP1) cells were treated with 1 M TSA or vehicle for 4 hours. Endogenous Ku70 was immunoprecipitated using goat anti-Ku70 antibodies or normal goat serum (NGS). Immunoprecipitates were separated by SDS-PAGE and probed with anti-Bax or anti-Ku70 antibodies. Input represents 20% of the sample used in immunoprecipitation. (B and C) Whole-cell extracts of N-type (IMR32 and SH-SY5Y in B; GOTO, SMS-KCN-69n, and LA1-55n in C) or S-type (SH-EP1 and SK-N-AS in B; LA1-5s, SH-310, and WSN in C) NB cells were separated by SDS-PAGE and probed with anti-CBP, anti-p300, anti-Ku70, and anti-Bax antibodies. -Tubulin was probed for equal loading.

NB cell lines. With respect to sensitivity to HDACIs (TSA, interaction between Bax and Ku70, as evidenced by the MS-275, and SAHA), they report that SHEP-SF cells, an failure to detect Bax in immune complexes precipitated with S-type line, are less responsive than N-type cells (CH-LA 90, anti-Ku70 antibody from CBP-transfected cells (Figure 3C). IMR-5, and SHSY-5Y), with a single exception in that SH- Furthermore, when CBP or p300 transfection was combined SY5Y cells were relatively resistant to SAHA. with TSA treatment, only CBP-transfected cells had increased HDAC inhibition can only increase Ku70 acetylation when sensitivity to TSA-induced death (Figure 3D, left panel). HAT activity capable of acetylating the corresponding sensi- These experiments were performed by harvesting cells that tive sites is present. Hence, these results showing that TSA remained viable after CBP transfection; these cells were then fails to disrupt Ku70–Bax binding in S-type cells could be treated with vehicle control, TSA, or doxorubicin (Dox). Unlike explained by the possibility that the cells lack adequate HAT the increased sensitivity to TSA seen in cells transfected with expression or activity. To test this possibility, we determined CBP, there was no change in the sensitivity of cells to Dox, the relative expression of Ku70, Bax, p300, and CBP, the which represents a cytotoxic agent that does not act by last of which is known to acetylate Ku70 [10]. As shown blocking HDAC activity (Figure 3D, right panel). Indeed, the in Figure 2, B and C, the N-type (IMR32 and SH-SY5Y in TSA dose response of CBP-transfected S-type cells was Figure 2B; GOTO, SMS-KCN-69n, and LA1-55n in Figure 2C) comparable to that of N-type cells (Figure 1A). Together, and S-type (SH-EP1 and SK-N-AS in Figure 2B;LA1-5s, these results show that the level of expression of CBP in SH-310, and WSN in Figure 2C) cell lines express nearly S-type cell lines can independently affect Ku70–Bax binding equal amounts of Ku70 and Bax; however, the levels of CBP and that increased CBP expression enhances cell death in and p300 are significantly lower in S-type cell lines than in response to TSA. N-type cell lines. To directly link Ku70 acetylation with CBP-induced Bax release, activation, and cell death, we took advantage of a Low-Level CBP Expression Limits the Responsiveness Ku70 mutant protein in which two acetylation-sensitive ly- of S-Type Cells to TSA Treatment sine residues (539 and 542) were replaced by arginines. This Next, we designed a series of experiments to determine mutant has been previously shown to block TSA-induced whether the level of expression of CBP or p300 governs apoptosis in N-type cell lines [4]. As shown in Figure 4A,a responsiveness to HDACIs in NB cell types. First, we tested single cotransfection using the CBP and Ku70 plasmids the hypothesis that increasing the level of CBP or p300 in resulted in the expression of target proteins. It was noted S-type cells would result in a more TSA-responsive pheno- that the level of CBP expression was lower when Ku70 type. FLAG-tagged CBP or p300 expression vectors were plasmid was also present, potentially due to transcription- transiently transfected into SH-EP1 cells by an electropora- level interference. As expected, transfection to overexpress tion method that results in a transfection efficiency of z 95% wild-type Ku70 or Ku70 lysine mutant did not significantly (data not shown). Twenty-four hours after transfection, affect viability (84% and 95% of controls; Figure 4B, lanes 5 SH-EP1 cell viability was assessed. Remarkably, although and 6, respectively), whereas CBP was toxic (52% viable; the expression levels of FLAG-tagged CBP and p300 were Figure 4B, lane 2). Cotransfecting wild-type Ku70 along with similar (Figure 3B), the transfection of CBP, but not p300, CBP resulted in slightly higher cell deaths (42% viability; had toxic effects; CBP reduced SH-EP1 viability to 30% of Figure 4B, lane 3) compared with transfecting CBP alone, vector-transfected control cells (Figure 3A). Mechanistically, whereas cotransfection of Ku70 mutant completely blocked CBP transfection, but not p300 transfection, disrupted the the toxic effects of CBP overexpression (81% viability;

Neoplasia . Vol. 9, No. 6, 2007 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al. 499

Figure 4B, lane 4). It is worth noting that the levels of CBP induced cell death. Using siRNA targeting CBP, we re- expression were nearly equal in cotransfections with wild- duced CBP protein expression in SH-SY5Y cells (up to 90%; type Ku70 and with Ku70 mutant (Figure 4A); therefore, Figure 5A). Depleting CBP in these cells significantly blocks differences in CBP expression do not account for differences the cytotoxic response observed after 24 or 48 hours of TSA in cell survival. We monitored the Bax–Ku70 complex fol- treatment (Figure 5B). These results indicate that CBP is a lowing each transfection. Consistent with effects on cell critical mediator of TSA-induced cell death. survival, transfection of CBP alone or CBP with wild-type Ku70 resulted in dissociation of Ku70–Bax (Figure 4C, lanes Bax Is Required for TSA-Induced Cell Death in NB Cells 6 and 7, respectively), whereas the complex remained intact We used siRNA knockdown of Bax and Ku70 in SH-SY5Y in cells cotransfected with CBP and Ku70 mutant (Figure 4C, cells (Figure 6A) to determine whether Bax is required for lane 8). These results provide evidence that CBP depends TSA-induced killing and to establish whether the balance in on critical acetylation-sensitive lysine residues in Ku70 for its the expression of Ku70 and Bax can affect viability. Consis- ability to disrupt the interaction between Ku70 and Bax in NB tent with the model developed above, reducing Bax expres- cells. However, these interpretations may be limited by the sion in SH-SY5Y cells protects cells from TSA-induced death unphysiologically high level of CBP expression introduced by (Figure 6C). Knocking down Bax expression, by itself, does transient transfection in these cells. not affect NB cell viability, whereas Ku70 knockdown alone led to cell death, even without TSA treatment (Figure 6B). High-Level CBP Expression Is Required for HDACI-Induced SH-SY5Y cells survived Ku70 knockdown, however, when Cell Death in N-Type Cells Bax was simultaneously knocked down and, as expected, Our data suggest that CBP may be required to acetylate death response to TSA was significantly reduced in double- Ku70 in NB cells. If so, we reasoned that depletion of CBP in knockdown cells (Figure 6C). These results show both that N-type cell lines should render cells less sensitive to TSA- the Ku70–Bax complex is a relevant target for therapeutic

Figure 3. Increasing CBP in S-type NB cells reduces cell viability and Bax binding to Ku70. SH-EP1 (S-type) cells were transfected with FLAG-tagged CBP, FLAG- tagged p300 expression vector, or empty vector. (A) Cell viability was determined by MTT assay 24 hours after transfection. Results are expressed as the percentage of viable cells compared with vehicle-treated controls (mean ± SD; n = 3). (B) The levels of CBP and p300 after transfection, as described above, are detected by immunoblotting with anti-FLAG, anti-CBP, and anti-p300 antibodies, as shown. -Tubulin was probed for equal loading. (C) SH-EP1 cells were transfected with FLAG- tagged CBP, FLAG-tagged p300 expression vectors, or empty vector. Twenty-four hours after transfection, Ku70 was immunoprecipitated using goat anti-Ku70 antibodies or NGS. Immunoprecipitates were separated by SDS-PAGE and probed with anti-Ku70 or anti-Bax antibodies. Input represents 20% of the sample used for immunoprecipitation. (D) SH-EP1 cells were transfected with FLAG-tagged CBP, FLAG-tagged p300 expression vector, or empty vector. Twenty-four hours after transfection, cells remaining viable were treated with various concentrations of TSA or Dox, as indicated, for an additional 24 hours, after which viability was determined by MTT assay. Results are expressed as the percentage of viable cells compared with vehicle-treated controls (mean ± SD; n = 3).

Neoplasia . Vol. 9, No. 6, 2007 500 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al.

Figure 4. Coexpression of a Ku70 acetylation mutant rescues CBP-induced reduction of cell viability in S-type NB cells. SH-EP1 (S-type) cells were transfected with FLAG-tagged CBP, FLAG-tagged wild-type Ku70, or FLAG-tagged Ku70 mutant (K539R/K542R) expression vectors, as indicated. The expression of transfected proteins is shown in (A) by immunoblotting using anti-FLAG antibodies. -Tubulin was probed for equal loading. (B) Twenty-four hours after TSA treatment, cell viability was determined by MTT assay. Results are expressed as the percentage of viable cells compared with vector alone (mean ± SD; n = 3). (C) Twenty-four hours after transfection, Ku70 was immunoprecipitated using goat anti-Ku70 antibodies or NGS. Immunoprecipitates were separated by SDS-PAGE and probed with anti-Ku70 or anti-Bax antibodies. Input represents 20% of the sample used in immunoprecipitation.

Figure 5. Reduction of CBP in N-type NB cells reduces TSA-induced cell death. (A) SH-SY5Y (N-type NB) cells were transfected with CBP-specific siRNA (5 or 10 nM) or scrambled siRNA. The control received no siRNA. Forty-eight hours after transfection, the level of CBP was determined by immunoblotting using anti- CBP antibodies. -Tubulin was probed for equal loading. (B) Forty-eight hours after siRNA transfection, the cells were treated with various concentrations of TSA for 24 or 48 hours, as indicated. Cell viability was then determined by MTT assay. Results are expressed as the percentage of viable cells compared with vector alone (mean ± SD; n = 3).

Neoplasia . Vol. 9, No. 6, 2007 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al. 501

Figure 6. Ku70 and Bax are critical for TSA-induced death of N-type NB cells. (A) SH-SY5Y cells were transfected with scrambled, Bax (10 nM), Ku70 (10 nM), or both Bax/Ku70 – targeted siRNA. The levels of Bax or Ku70 protein at 48 hours after transfection are shown in immunoblots using Bax-specific or Ku70-specific antibodies. -Tubulin was probed for equal loading. (B) Forty-eight hours after siRNA transfection, the viability of Bax, Ku70, or Ku70/Bax knockdown in SH-SY5Y cells was determined by trypan blue exclusion assay. The results are expressed as percentages of viable cells relative to cells transfected with scrambled siRNA (mean ± SD; n = 3). (C) Forty-eight hours after siRNA transfection, Bax or Ku70/Bax knockdown in SH-SY5Y cells was treated with various concentrations of TSA, as shown. The viability of cells was determined by MTT assay 24 or 48 hours after TSA treatment by comparison to untreated SH-SY5Y cells. The results are expressed as mean ± SD (n = 3).

strategies to decrease NB survival and that Bax expression 3.65), although there is a smaller difference in the expression is required for TSA-induced cell death. of Ku70 or Bax between neuroblastic and stromal regions (neuroblastic/stromal ratios are 2.21 and 1.36, respectively). Proteins Required for HDACI-Induced Cell Death Are Semiquantitative analyses of CBP and p300 expression Expressed in Human NB Tumors levels in neuroblastic foci reveal that > 90% of tumors dem- Next, we determined the protein-level expression of Bax, onstrate expression that reaches the upper half of the range Ku70, and CBP in human NB tumors to assess the potential (levels 2 and 3) (Figure 7B). In contrast, CBP and p300 of these proteins to be developed as biomarkers to select expression in stromal regions reaches the upper range in tumors that are most likely to respond to HDACIs. To do only 20% to 35% of tumors and is barely detectable in > 35% this, we constructed NB tumor tissue microarrays and used of stromal regions. Thus, CBP and p300 are differentially immunohistochemical detection methods to semiquantita- expressed between neuroblastic and stromal cell regions in tively determine the expression of Bax, Ku70, CBP, and tumors, mirroring the differential expression observed be- p300 (Figure 7A). tween N-type and S-type NB cell lines. In addition to dem- We found that every NB tumor had detectable levels of onstrating the presence of these proteins in all NB tumor these proteins. This suggests that the critical components of specimens, these data show a range for the expression for the TSA response mechanism defined in vitro are present each protein across tumors that varies according to neuro- in tumors. It is interesting to note that CBP and p300 are blastic and stromal histology. The differences in expression expressed at higher levels in neuroblastic regions of tumors observed across this small set of NB tumors support the pos- compared to the stroma (the CBP neuroblast/stroma ratio is sibility of detecting correlations between the expression of

Neoplasia . Vol. 9, No. 6, 2007 502 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al. levels of CBP, Ku70, or Bax and clinical variables such as which CBP, typically considered a nuclear protein, may be responses to HDACI treatment in a future translational study. active in the cytoplasm where it acetylates Ku70. CBP and the related molecule p300 have previously been detected in the cytoplasm of other cells. For example, CPB and p300 are Discussion cytoplasmic in mouse oocytes isolated from primordial fol- Our results support a model in which the association of Bax licles, but as follicles develop, oocyte CBP and p300 become with Ku70 in NB cells defines a target for TSA and in which progressively localized in the nucleus [13]. Cohen et al. [10] response to TSA treatment is regulated by CBP. This model have shown that, after HDACI treatment, CBP translocates further predicts that the sensitivity of NB tumor cells to the from the nucleus to the cytoplasm in 293Tcells. In NB, HeLa, apoptotic effects of certain HDACIs depends on the three and HEK 293 cells, we have recently demonstrated that CBP components: Ku70, Bax, and CBP. We have tested this and p300 are detected in both the cytoplasm and the nucleus model in S-type NB cell lines in which the level of CBP is (manuscript in preparation). These results provide for the low. Overexpression of CBP in S-type cell lines favors the possibility that cytoplasmic CBP acetylates Ku70. It is also dissociation of Bax from Ku70, increasing the sensitivity of possible that the interaction between CBP and Ku70 occurs these cells to TSA. A Ku70 acetylation site mutant, which we in the nucleus and that shuttling of Ku70 between the nu- have previously shown, blocks the apoptotic effects of HDA- cleus and the cytoplasm allows the acetylation that occurs in CIs in N-type NB cell lines and also blocks CBP-induced cell the nucleus to ultimately affect the binding equilibrium of death in S-type NB cells. Conversely, reducing CBP expres- cytoplasmic Ku70 and Bax. Ongoing work is directed at sion in N-type NB cell lines significantly reduces the cytotoxic answering this question. effects of TSA, further establishing that CBP is a factor determining HDACI responsiveness. The results of experi- CBP, But Not p300, Reduces the Viability of S-Type ments in which Bax and Ku70 expression was modulated NB Cells not only support the importance of the interaction between Because CBP and p300 are highly homologous (61% Bax and Ku70 in maintaining NB cell viability but also dem- identical at the amino acid level overall, and 95% similar in onstrate that TSA-induced cell death in NB cells depends on functional domains), it is surprising that only the expression of these proteins because cells lacking both Ku70 and Bax are CBP, and not p300, is cytotoxic to S-type NB cells (Figure 3). insensitive to HDACI-induced killing. CBP and p300 are functionally equivalent in terms of their interactions with many cellular proto-oncogenes and a wide CBP Mediates Cell Death By Affecting Ku70–Bax Binding variety of transcription factors [14]. Mutations of CBP and in the Cytoplasm p300 in humans are both associated with the Rubinstein-Taybi Our results suggest that CBP acetylates Ku70 as part of a syndrome, a developmental disorder characterized by mental mechanism leading to the disruption of cytoplasmic Ku70– retardation, digital abnormalities, and an increased incidence Bax complexes. These data are consistent with a model in of certain types of tumors, including NB [15,16]. Nevertheless,

Figure 7. Expression of Bax, Ku70, and CBP in a tissue microarray of neuroblastic tumors. (A) Cores of formalin-fixed paraffin-embedded neuroblastic tumors were stained using peroxidase-conjugated anti-Bax, anti-Ku70, or anti-CBP antibodies (brown signals). The slides were then counterstained with hematoxylin (blue background). (B) Fraction of N-type or S-type cells that express different staining intensities (levels, 0 – 3+) of CBP, p300, Ku70, or Bax in our NB tissue microarray studies.

Neoplasia . Vol. 9, No. 6, 2007 CREB-Binding Protein as Mediator of Neuroblastoma Cell Death Subramanian et al. 503 several examples provide evidence that the functions of CBP cancer therapy: is transcription the primary target? Cancer Cell 4 (1), 13 – 18. and p300 do not completely overlap. In F9 cells, retinoic acid– [4] Subramanian C, Opipari AW Jr, Bian X, Castle VP, and Kwok RP induced differentiation is dependent on p300, but not CBP, (2005). Ku70 acetylation mediates neuroblastoma cell death induced because expressing a p300-specific hammerhead ribozyme by histone deacetylase inhibitors. Proc Natl Acad Sci USA 102 (13), 4842 – 4847. capable of cleaving p300 messenger RNA, but not CBP, [5] Sawada M, Sun W, Hayes P, Leskov K, Boothman DA, and Matsuyama makes these cells resistant to retinoic acid–induced differen- S (2003). Ku70 suppresses the apoptotic translocation of Bax to mito- tiation [17]. Although p300, and not CBP, is crucial for proper chondria. Nat Cell Biol 5 (4), 320 – 329. [6] Mimori T, Hardin JA, and Steitz JA (1986). Characterization of the DNA- hematopoietic differentiation [18], only CBP is essential for binding protein antigen Ku recognized by autoantibodies from patients hematopoietic stem cell self-renewal. The results in the cur- with rheumatic disorders. J Biol Chem 261 (5), 2274 – 2278. rent study show that CBP and p300 are not functionally [7] Rathmell WK and Chu G (1994). Involvement of the Ku autoantigen in the cellular response to DNA double-strand breaks. Proc Natl Acad Sci equivalent in NB cells and suggest that this may result from USA 91 (16), 7623 – 7627. differences in substrate specificity for Ku70 acetylation. The [8] Dvir A, Peterson SR, Knuth MW, Lu H, and Dynan WS (1992). Ku precise mechanism that determines CBP and p300 substrate autoantigen is the regulatory component of a template-associated protein kinase that phosphorylates RNA polymerase II. Proc Natl Acad Sci specificity remains an area of intense research interest. USA 89 (24), 11920 – 11924. [9] Gottlieb TM and Jackson SP (1993). The DNA-dependent protein ki- Clinical Relevance nase: requirement for DNA ends and association with Ku antigen. Cell 72 (1), 131 – 142. NB tumors are comprised of neuroblastic and stromal [10] Cohen HY, Lavu S, Bitterman KJ, Hekking B, Imahiyerobo TA, Miller C, cells [19,20]. Differences in the proportion of neuroblastic Frye R, Ploegh H, Kessler BM, and Sinclair DA (2004). Acetylation of and schwannian stromal cells present in neuroblastic tu- the C terminus of Ku70 by CBP and PCAF controls Bax-mediated apoptosis. Mol Cell 13 (5), 627 – 638. mors, which include NB, ganglioneuroblastoma, and ganglio- [11] Cohen HY, Miller C, Bitterman KJ, Wall NR, Hekking B, Kessler B, neuroma, are one of the important criteria used to establish Howitz KT, Gorospe M, de Cabo R, and Sinclair DA (2004). Calorie clinical prognosis. Tumors that are composed primarily of restriction promotes mammalian cell survival by inducing the SIRT1 deacetylase. Science 305 (5682), 390 – 392. neuroblasts and have low numbers of schwannian stromal [12] Furchert SE, Lanvers-Kaminsky C, Juurgens H, Jung M, Loidl A, and cells are more aggressive and are usually the most difficult Fruhwald MC (2007). Inhibitors of histone deacetylases as potential to cure [21,22]. Our analysis of a panel of neuroblastic tu- therapeutic tools for high-risk embryonal tumors of the nervous system of childhood. Int J Cancer 120 (8), 1787 – 1794. mors identifies a difference in the level of CBP expression [13] Kwok RP, Liu XT, and Smith GD (2006). Distribution of co-activators between neuroblastic and schwannian stromal cells. Spe- CBP and p300 during mouse oocyte and embryo development. Mol cifically, neuroblastic cells express three-fold more of this Reprod Dev 73 (7), 885 – 894. [14] Goodman RH and Smolik S (2000). CBP/p300 in cell growth, trans- protein. Because TSA is more active against N-type cells formation, and development. Genes Dev 14 (13), 1553 – 1577. compared to S-type cells in culture and because its killing [15] Petrij F, Giles RH, Dauwerse HG, Saris JJ, Hennekam RC, Masuno M, depends on adequate CBP expression, we predict that Tommerup N, van Ommen GJ, Goodman RH, Peters DJ, et al. (1995). Rubinstein-Taybi syndrome caused by mutations in the transcriptional HDACIs will preferentially kill neuroblastic cells within these co-activator CBP [see comments]. Nature 376 (6538), 348 – 351. tumors, which may mean that tumors with higher neuro- [16] Roelfsema JH, White SJ, Ariyurek Y, Bartholdi D, Niedrist D, Papadia blastic content (NBs) will respond better to these agents than F, Bacino CA, den Dunnen JT, van Ommen GJ, Breuning MH, et al. (2005). Genetic heterogeneity in Rubinstein-Taybi syndrome: muta- stromal-predominant ones (ganglioneuromas). Further ex- tions in both the CBP and EP300 genes cause disease. Am J Hum periments and clinical trials that validate the safety and ef- Genet 76 (4), 572 – 580. ficacy of HDACIs in patients with neuroblastic tumors may [17] Ugai H, Uchida K, Kawasaki H, and Yokoyama KK (1999). The co- activators p300 and CBP have different functions during the differentia- consider evaluating whether clinical variables, including tion of F9 cells. J Mol Med 77 (6), 481 – 494. responses to this therapeutic class, correlate with the ex- [18] Rebel VI, Kung AL, Tanner EA, Yang H, Bronson RT, and Livingston pression of Bax, Ku70, and CBP. DM (2002). Distinct roles for CREB-binding protein and p300 in hematopoietic stem cell self-renewal. Proc Natl Acad Sci USA 99 (23), 14789 – 14794. [19] Valent A, Benard J, Venuat AM, Silva J, Duverger A, Duarte N, Hartmann Acknowledgements O, Spengler BA, and Bernheim A (1999). Phenotypic and genotypic di- versity of human neuroblastoma studied in three IGR cell line models We would like to thank Yihong Liu for technical support and derived from bone marrow metastases. Cancer Genet Cytogenet 112 Nicole Gross (Pediatric Oncology Research, University Hos- (2), 124 – 129. pital CHUV, Switzerland) for neuroblastic and stromal NB [20] Mora J, Cheung NK, Juan G, Illei P, Cheung I, Akram M, Chi S, Ladanyi M, Cordon-Cardo C, and Gerald WL (2001). Neuroblastic cell lines. and Schwannian stromal cells of neuroblastoma are derived from a tumoral progenitor cell. Cancer Res 61 (18), 6892 – 6898. [21] Shimada H, Ambros IM, Dehner LP, Hata J, Joshi VV, Roald B, Stram DO, Gerbing RB, Lukens JN, Matthay KK, et al. (1999). The Interna- References tional Neuroblastoma Pathology Classification (the Shimada system). [1] Sterner DE and Berger SL (2000). Acetylation of histones and Cancer 86 (2), 364 – 372. transcription-related factors. Microbiol Mol Biol Rev 64 (2), 435 – 459. [22] Shimada H, Umehara S, Monobe Y, Hachitanda Y, Nakagawa A, Goto [2] Marks PA, Richon VM, and Rifkind RA (2000). Histone deacetylase S, Gerbing RB, Stram DO, Lukens JN, and Matthay KK (2001). Interna- inhibitors: inducers of differentiation or apoptosis of transformed cells. tional neuroblastoma pathology classification for prognostic evaluation J Natl Cancer Inst 92 (15), 1210 – 1216. of patients with peripheral neuroblastic tumors: a report from the Child- [3] Johnstone RW and Licht JD (2003). Histone deacetylase inhibitors in ren’s Cancer Group. Cancer 92 (9), 2451 – 2461.

Neoplasia . Vol. 9, No. 6, 2007